|updated Friday, December 17, 2004||By Philip G. Stewart, pstewart+atsymbol+gwi+dot+net|
The Nature and Use of This Web Page
Overview of the project
What the support circuitry must do
Support chips and system organization
Implementing the support circuitry
The Nature and Use of This Web Page
Here we document our reverse-engineering of a video system to display standard NTSC composite video, e.g. from a camcorder or VCR, on a postage-stamp-sized liquid crystal display (LCD), the Matsushita MCL0712A03.
This LCD, sold as surplus with its matching backlight by BG Micro,
Halted Specialties, and others, comes with data sheets that give
inadequate details on the driver circuitry it needs to make it work.
Our reverse-engineering effort fills in the blanks in this
documentation. The end product of our effort is a system that can
display either standard NTSC composite video or computer-style RGB
video with NTSC timing.
This page, 'The Care and Feeding of the Cybermaxx LCD,' presents our results to date. It distills the knowledge gleaned in all our experimentation into one document. It will prepare you to participate fully in the reverse-engineering process, but it will not tell you precisely what is needed to make the LCDs work (as of this writing) because some questions remain to be cleared up. You can think of this document as the predecessor to the kind of construction manual you get with a hobby kit; once this document is finalized, that one can be written.
The present document is organized as follows (excluding the section you are reading now):
(1) Overview. This section gives some background on the project; explains why the reverse engineering effort has been necessary; and sketches an outline of the reverse engineering process we have used. While each section is useful, the reverse engineering outline at the end is most important for understanding our design process.
(2) What the support circuitry must do. This gives an outline of what needs to happen in the video system so that the LCD works.
(3) Support chips and system organization. This section introduces the Application-Specific Integrated Circuits (ASICs) used in the Cybermaxx to feed the LCDs the signals they need. We illustrate their functions with a system block diagram, which serves as a bridge from the abstract question of what signals are needed to the matter of how in practice they are generated.
(4) Implementing the support circuitry. This section gets down to the practical details of how the system is designed and put together.
We give concrete design details on our PC boards page for now, pending a fully polished design. Other pages at this site, for example our schematics page and data sheets pages, give extensive supplementary information which is very important to consider alongside this document, e.g. timing diagrams that are easiest for us to present as Adobe Acrobat PDF documents. More practical information appears on our parts sources page. For those who are interested in placing this design information in the broader context of a virtual reality system, there is a great deal of information available in links from our resources page.
Where the LCDs come from: The Cybermaxx Virtual Reality Helmet
Missing Documentation Entails Reverse Engineering Effort
Overview of the Reverse Engineering Process
1. Where the LCDs come from: The Cybermaxx Virtual Reality Helmet.
A virtual reality (VR) helmet is designed to immerse the one who wears
it in a computer-generated world, presenting pictures and sounds that
follow one's head movements as if one is exploring an actual 3D
Two versions of the Cybermaxx were sold, one with two 505x230 displays, for a total of 116,150 pixels each, and a second model, with an improved 180,000 pixel display. When the Victormaxx company got out of the consumer electronics business, it sold off its stock of parts, which ended up in the catalogs and shops of a number of surplus electronics retailers, including Halted Specialties, Inc., in Silicon Valley, BG Micro, and one store in Toronto.
Among these parts were the 505x230 color graphics LCDs (liquid crystal displays), designated MCL0712A03 by Matsushita, their manufacturer. These LCDs are the subject of our design efforts.
The LCDs themselves lack driver circuitry. In order for them to display video, we need to provide them with the proper power supply voltages and signals. The surplus dealers that sold them (BG Micro, Halted Specialties, et al.) provided photocopies of a data sheet with them that appeared to be intended for use by a test facility, emphasizing reliability measures and thermal tolerances. Absent was any reference to the support circuitry, or documentation of how to set it up.
2. A Reverse Engineering Effort Has Been Necessary.
A reverse engineering effort has thus been necessary to figure out how to get the Matsushita MCL0712A03 LCD working. A very compressed history of this effort follows:
Inadequate documentation from Matsushita and inadequate assistance from their subsidiary Panasonic have added difficulty to the project. Difficulties in obtaining the driver circuitry from Panasonic at first led us to believe we would have to fully reverse-engineer the driver integrated circuits. Difficulty in using the integrated circuits we finally obtained from Panasonic arose when they refused to provide data sheets on how to apply the devices. The reason they gave was that they did not want to encourage reverse engineering of the I.C.s by publishing their specifications; in effect we were almost forced to do precisely this, because Panasonic would not tell us how to use the circuits we bought from them! Correspondents in Belgium and Silicon Valley, however, managed to obtain reasonably detailed data sheets for the I.C.s, allowing us to simplify our project immensely. While the data sheets lack many of the amenities one finds in integrated circuit documentation (such as input and output equivalent-circuit diagrams and detailed application information), they have made it unnecessary for us to create our own video driver chips.
Sharp Microelectronics, manufacturer of the other video chip in the Cybermaxx, has by contrast been immensely helpful, providing top-quality, fully informative, professionally prepared data sheets in a timely, friendly, and fuss-free fashion. The quality of Sharp's documentation has aided our project greatly, making many things clear that are not so clear in the Panasonic documents.
Panasonic's detailed camcorder repair manuals, however, have been of great help, filling in blanks where the scanty application information for their I.C.s had left us uncertain of what do to. Despite inconsistencies in the information they give, they have been very helpful.
The assiduous and skilled work of Nate Caine, who reverse-engineered the Cybermaxx video system himself, has been crucial to our efforts as well. His discussion of this work, originally hosted at Halted Specialties' Web site, and now hosted by virtual reality experimenter Kevin Mellott, is essential reading for anyone working with the LCDs or other parts of the Cybermaxx system. Unfortunately we have not had the benefit of his schematics, which are not in a machine-readable form and would require extensive rework to be made so.
The most recent work on the project has been conducted as part of the design program of Delectra Jouet, led by Julie S. Porter.
3. Overview of the Reverse Engineering Process.
Our objective is to replicate partially known system designs, using the same integrated circuits, but without complete documentation on either the constituent I.C.s or of the whole design. We draw on varied information sources to fill in gaps in our knowledge, as we will now describe.
Our LCDs display NTSC video (NTSC is the North American television standard). But it is not "straight" NTSC video; it is video with NTSC timing. The LCDs need circuitry to convert an incoming stream of standard NTSC video (such as a VCR puts out) to a form they can use. For this, they need a video converter system, and a separate system to convert video timing signals into the logic control signals that run the LCD.
We must copy existing examples of this circuitry to get the LCDs working, or else build an entire video system from scratch. In either case it is best that we look at working systems using LCDs like ours before setting to work, because we lack sufficient documentation on them to know what their full functional requirements are. We use what we have learned about the Cybermaxx and related systems to fill in where our knowledge is lacking.
Finding a model to work from in lieu of the Cybermaxx
Victormaxx used existing camcorder technology for the Cybermaxx's video display. Each of the Cybermaxx's two LCDs (one for each eye) is essentially a 'color electronic viewfinder' from a camcorder, with some adaptations built in to integrate them into a stereoptical virtual reality viewer. Although the LCD's part number doesn't show up in the Panasonic parts database (according to the Panasonic Services Corporation's telephone people), it is of the same 505x230 configuration as the LCDs in some Panasonic camcorder viewfinders. It is manufactured by Matsushita (Panasonic's parent company) and is driven by the same proprietary Matsushita timing control chips that these camcorders use. In the Cybermaxx it gets its RGB video from a chip also found in Panasonic camcorder electronic viewfinders, the Sharp Microelectronics IR3Y05-- a chip designed specially for electronic viewfinders. For these reasons, when it proved impossible to locate actual Cybermaxx schematics, I looked to the LCD's natural habitat, the camcorder, to see how it was used there.
Panasonic's camcorder repair manuals afford good examples of working LCD systems, with detailed schematics-- something which (as noted) we lack for the Cybermaxx. For this reason our designs will very closely follow the designs of a camcorder that uses circuitry similar to the Cybermaxx; however, we will introduce modifications when the camcorder circuitry fails to match the specifications supplied for the LCD, or when errors in the camcorder schematics or contradictions between them and the integrated circuit data sheets arise. Thus, in the section below on 'Implementing the Support Circuitry,' we will discuss how the camcorder designs implement the support chips, and how we may modify this design to suit the MCL0712A03 LCD. Where there is uncertainty or disagreement between I.C. data sheets and camcorder schematics, we apprise the reader of this, so that s/he can take part fully in the design process, and pick up where we leave off as necessary.
We supply supplementary design analysis elsewhere on this web site, on our schematics page, and will refer to it as necessary in this document. Most of this analysis deals with differences between the camcorder schematics, such as differences in the timing driver chip MN83803Ax's sync input filtering networks, and in its Voltage Controlled Oscillator (VCO) external components. Some details that do not appear to have relevance to driver system functionality are, naturally, left out in this document.
What The Support Circuitry Must Do
(1) It must provide an RGB analog video signal the LCD can use.
It must do several things to accomplish this:
(1.1) It must convert NTSC to RGB video.
Standard NTSC video travels on one wire (so it can be broadcast efficiently) but the LCD has three video input wires (for red, green, and blue, or 'RGB' video). Another term for one-wire video is 'composite' video; its three-wire counterpart is 'component' video.
With each successive video frame, it must switch the polarity of this
RGB signal, so that it alternates from positive to negative and back.
To make a long story short, this keeps the LCD from burning out thanks to an idiosyncracy of liquid crystal display design. (you will find a good explanation of it in Nate Caine's discussion of the Cybermaxx on Kevin Mellott's Web site, as well. See "RGB Inversion" near bottom of page.).
Panasonic camcorders and the Cybermaxx use a timing chip that counts lines rather than video frames, so in the devices we look at, alternation happens on successive video lines. But see our discussion of the Cybermaxx's apparently faulty implementation of this, following from Nate Caine's notice of it.
(1.3) It must shift this RGB signal's level so that it averages 7 volts above system ground.
(2) It must convert NTSC's simple timing signals into a form the LCD can use.
The LCD needs an elaborate set of timing signals that are delivered to it in step with the NTSC sync.
(3) In addition to NTSC-to-RGB conversion and timing, the circuit must provide:
(3.1) A timing signal for the LCD's backlight, and
(3.2) Proper power supply voltages for the LCD.
Support Chips and System Organization
The camcorders we take as our models use two integrated circuits to convert the NTSC input into a signal the LCDs can use:
(1) The Sharp IR3y05y or IR3y05
This is the video processing chip. It does all the conversions listed above except (2), the timing conversion, and (3.1), the production of a backlight control pulse. To prepare the timing conversion, it separates the synchronization signal ('sync' for short) from the input and feeds it to a second integrated circuit, the MN83803Ax (MN83803A, MN83803AK, MN83803AK/K, or MN83803AL), which transforms it into the LCD's timing signals. More details on what this chip needs to run are provided below.
(2) The Matsushita (Panasonic) MN83803Ax (MN83803A, MN83803AK, MN83803AK/K, or MN83803AL)
This device controls the LCD's internal registers. In addition to running the LCD, it runs the backlight that illuminates the LCD ((3.1), above), and tells the Sharp IR3y05y when lines begin and end, so it can invert the video signal, as described above, on alternating lines. More information on what this chip needs to run is given below.
The schematics for the camcorders we base our design on, the Panasonic PV-950-B and PV-950-B/K, can be boiled down to the following simple block diagram:
The essentials of the design can be gleaned from this. The signal entering the system at the upper left is standard NTSC composite video.
Implementing The Support Circuitry
We can now consider the LCD's needs in more detail. First we will consider the direct inputs to the LCD; then, what is needed to run its support chips and backlight; then construction details and some questions concerning system integration..
(5) Physical design
(5.1) Surface-mount techniques
(6) System Integration
(6.1) I.C. interchangeability questions
(6.2) Supply voltage questions
(1) LCD Inputs
What the LCD needs to make it work:
(1.1) RGB analog video input, timed according to the NTSC standard.
(1.1.0) What is NTSC video?
To understand how the LCD works, it's important for us to understand the kind of signal it's meant to display. Since it is designed for use in camcorders, it displays television video; specifically, it displays NTSC video, the North American standard specified by the National Television Standards Committee (more recently this standard has been renamed RS-170; color NTSC video is specified by a revised standard, RS-170A; the phrase 'NTSC video' continues to be widely used to refer to them.).
While NTSC video is much more complicated than we can describe here, a brief overview of its timing will help us to understand the LCD's workings.
NTSC scanning is patterned exactly like reading a book printed in English; just as we read left to right for each line on a page, and from the top line on the page to the bottom, NTSC video signals are delivered dot by dot, left to right, and line by line, top to bottom. A single full scan of the video screen, analogous to a single frame in a movie, is called a frame. Thirty frames of video are displayed per second, but each frame is divided into two 'fields,' one displaying odd lines, the other even. This scheme, called 'interlacing,' gives sixty fields per second.
At the end of each horizontal line there is a brief interval when a television screen is blank, allowing the TV's electron beam to retrace its path to start the next line without making a visible streak on the screen. A similar thing happens at the end of a video field, so that the electron beam can retrace its path to the upper left corner for the start of the next field, without making a diagonal streak from the lower right to the upper left corner. An LCD doesn't need this blanking interval to work, but since ours displays video designed for televisions' cathode ray tubes (CRTs), it must take its video input in the same format as CRTs do. Each line of video is followed by a horizontal blanking interval, and each field is followed by a (longer) vertical blanking interval. The LCD's timing of lines and blanking intervals adheres to NTSC's timing.
The synchronization (sync) pulse in NTSC video is delivered during the blanking interval. It is necessary-- or at least most convenient-- to put it here, so that it won't interfere with the delivery of the video signal itself. Alternating sync with the video signal like this enables the signal to be delivered serially, on one transmission line, making it suitable for broadcasting, and easy to circulate on simple coaxial cables.
Here (below) is an oscilloscope picture centering on one horizontal line of NTSC composite video. It shows some of NTSC's timing and voltage parameters.
You should familiarize yourself with the NTSC specification, and with the LCD's data pertaining to it, to understand the LCD's data sheets. While not difficult in principal, learning about NTSC does entail mastery of quite a few more or less arcane details. Knowing the NTSC standard allows one to understand the kind of information the LCD timing system takes as its input, and thus gives some idea of the kinds of transformations this timing system will perform in generating the LCD control signals. Keep in mind that there is some flexibility built into the standard. The MN83803A(K) data sheet (in PDF form on this web site's 'data sheets' page') is also useful for this purpose, containing detailed timing charts.
(1.1.1) Signal inversion
The RGB video must alternate from positive to negative (with respect to the RGB center (i.e. average) voltage) with each video frame, or else the time-averaged DC voltage difference from the center voltage will degrade and eventually destroy the LCD.
The device that makes this alternation happen is called the 'inverter' (more properly an 'optional inverter' or 'alternating inverter'). As you can see in the system diagram in the previous section, the IR3y05 video processing chip, by Sharp, has the necessary inverter built in, controlled by its Frame Rate Pulse (FRP) input pin.
As Nate Caine explains in his Cybermaxx page, the Cybermaxx switches the RGB voltage polarity with each line, rather than with each frame. This follows from the device's use of the MN83803AK timing chip, which provides a signal that alternates line by line, rather than frame by frame (this is the "HT" (presumably "Horizontal Timing") signal). It may just be more convenient to use this signal to control the IR3y05's voltage inverter alternation than it would be to add a device that distinguishes odd from even frames.
In Nate Caine's discussion, he points out that the Cybermaxx implements its stereoscopic imaging scheme in a way that appears to allow a destructive net (time-averaged) DC voltage to build up in the LCD, since its signal alternates between the left LCD and the right, rather than line by line in the same LCD. Lacking Cybermaxx schematics, we take his word for this. Given an odd number of lines per frame, we can presume that in a single-LCD implementation, the line-by-line alternation that the MN83803Ax's "HT" signal facilitates will stand in for a frame-by-frame alternation such as the IR3y05 is designed to provide.
Since this issue appears to arise at least partly from the mix of chips used for the LCD system (the IR3y05 and MN83803Ax), it is properly a system integration issue, but for now we leave treatment of it to the present section only, lacking additional noteworthy details.
(1.1.2) DC offset of video center voltage
According to the LCD specification, the RGB video signal must be offset from system ground by +7.0 volts DC. Since the magnitude of the signal displayed at each pixel of the LCD is a function of the absolute value of how much the RGB signal departs from the center voltage, you get the same pixel displayed for +12 volts (+7 +5) as you do for +2 volts (+7 -5) input.
Controlling the Video Center Voltage:
Video signal center voltage for the MCL0712A03 is 7.0 volts (typical), or from 6.8 volts (min) to 7.7 volts (max). By default, the IR3y05 video processing chip sets the video center voltage to Vcc2 / 2. An external voltage reference can be applied to Pin 28 to set it to a value independent of the power supply (cf. IR3y05 Supply Voltage and the RGB Video Center Voltage, in section (6.2.1) of 'Power Supply Considerations', and Video Center Voltage (DC Offset) Control, section (3.2), below.)
Empirical testing shows that if the LCD's RGB center voltage (DC offset) sags below 7 volts, you can set the COM voltage lower as well and still get the LCD to work. Whether this damages the LCD in the long run or not is unclear; it will at least cut dynamic range from the RGB signal if its DC offset departs from 7 volts.
(1.2) Horizontal and Vertical Timing Signals.
These signals are needed to control when the RGB values from the analog inputs are scanned onto lines on the screen.
It isn't crucial to know exactly how these signals work, unless you are reverse engineering the MN83803AK or one of its sister chips (a subproject we considered when it wasn't clear we could get the MN83803AK chip from Panasonic, and which may still make an interesting exercise).
It still may be useful, even if not crucial, to understand the timing signals and the LCD's control logic, in case it is necessary to troubleshoot the circuit you build, or to make sure it is working by feeding its output waveforms to an oscilloscope or logic analyzer.
Consult Nate Caine's article on the workings of the Cybermaxx VR helm for a detailed diagram of the logic circuitry these lines feed in the LCD. Also see his verbal description of LCD timings on our site, for some help in figuring out what is going on. The MN83803AK manual will also show you detailed information about the specifications of the timing signals. With a little work, you can see how almost all of the timing relates to the timing of the NTSC video signals the LCD will display.
(1.2.1) The MN83803Ax series of LSI timing chips is produced by Panasonic exclusively to drive their LCDs. Each of these chips provides the horizontal and vertical timing pulses the LCD needs, given simply an NTSC sync input. These chips are considered in detail in section (2), below.
(1.2.2) These chips must be set for the appropriate timing mode for the particular LCD you are using, since they are designed to support several different types of LCD. See section (2.2), below, for details.
(1.2.3) The important thing is to get the signals the MN83803Ax chips will produce while running freely to lock onto the sync of the video that you're feeding to the LCD, so that the horizontal and vertical control pulses they are sending to the LCD are properly timed. (When we say 'lock on to the sync,' we mean 'become entrained to it,' or 'fall into proper synchrony with it', 'get in step with it.' The usual way to do this electronically is with a phase-locked loop (PLL)—This is a closed-loop control circuit that compares the phase of an oscillator's signal with a reference signal (e.g. incoming NTSC sync pulses) and minimizes the timing difference between them. The MN83803Ax's voltage-controlled oscillator (VCO) forms part of a PLL.)
(1.3) DC power supply
The LCD data sheets specify two power supply lines (besides ground, aka VSS): VDD and COM.
The main supply is +18 volts.
LCcom (or COM, or Vcom) is a DC voltage the LCD needs to make its transistors work, to make a long story short. As noted in the diagram above, it is directly related to the RGB center voltage; it must fall within the range of Video Center Voltage -1.9 volts to Video Center Voltage - 1.5 volts, with Video Center Voltage - 1.7 volts the typical setting. So for an RGB Video Center Voltage of 7.0 volts, LCcom would be set typically to 5.3 volts.
Empirical testing shows that if the LCD's RGB center voltage (DC offset) sags below 7 volts, you can set the COM voltage lower as well and still get the LCD to work. Whether this damages the LCD in the long run or not is unclear; it will at least cut dynamic range from the RGB signal if its DC offset departs from 7 volts.
You will need to look at
the voltage divider network feeding LCcom to the LCD, and make sure
that it is appropriate, given your system power supply and RGB center
voltage. Voltage dividers are easy to design. Just don't use the values
straight off the camcorder schematics, because these are for a
different LCD than we are presently using.
(2) MN83803Ax series timing chip
You can see a block diagram of the MN83803AK's internal workings at our schematics page or by clicking here (130K download). By itself, this block diagram doesn't seem as illuminating as it could be, but as part of the schematic diagram it appears in, it is good to look at alongside the present discussion. So are the MN83803AK data sheets (Adobe Acrobat PDF format, 640K download). You will notice, if you look at the block diagram referred to above (same 130K download), that pin 7 ('SYNCOUT') is not only not connected, but its internal connections are not shown. In the PV950-B/PV950B-K camcorder repair manual, there is a block diagram of the MN83803AL separate from the schematic diagram of the EVF, but I do not have a scan of it yet.
(2.1) Sync Input to the MN83803Ax series Chips
We are still working on fully specifying what these chips really need for a sync input, and how to pretreat it before it goes into the chip. Julie S. Porter has experimented with this, and come up with some useful results.
Click here to go directly to Julie's correspondence describing her test setup, procedures, and results.
(2.1.1) The MN83803Ax chips take their NTSC timing at Pin 6 ('SYNC/VIDEO'). The timing signal may be negative-going (in 'VIDEO' mode), consistent with the NTSC standard (-0.286 volts sync tip), or positive-going (in 'SYNC' mode). The timing chart for the 'measurement circuit' in the MN83803AK data sheets suggests that VIDEO mode can take straight NTSC composite video, capacitively coupled into pin 6, to get its sync signal. The IR3y05 video processing chip delivers a positive-going (max 0.6 volt) sync from its sync separator circuit in the camcorder manuals we've obtained-- and so in these, pin 6 of the MN83803Ax chips is set to 'SYNC' mode (positive-going sync input).
(2.1.2) S/V (Pin 41) on the MN83803Ax chips toggles Pin 6 between SYNC and VIDEO mode.
(2.1.3) Since Julie had to apply quite a bit more negative voltage to Pin 6 (in VIDEO, i.e. negative-going sync mode) than NTSC specifies, there is some question as to the proper setup of the signal treatment network feeding Pin 6, and the exact specifications of these chips' SYNC/VIDEO inputs. Only incomplete information is provided in the available data sheets and manuals.
(2.1.4) Our most recent design closely follows the camcorder schematic for Panasonic's PV-950B, so we use the Sharp IR3y05 (or IR3y05y) for video processing and sync separation. Since, as noted, the IR3y05 delivers a positive-going sync pulse 'ON' state, we set the MN83803AL's Pin 6 to 'SYNC' mode (so we hold Pin 41, S/V, 'High').
(2.2) Logic Control Signals for MN83803Ax Chips
Logic pins held high or low determine what LCD configuration the timing chip works for, as well as the functions of various pins, most notably Pin 6 (SYNC/VIDEO) for our purposes. If you're switching these, you may want to hold them high or low via a resistor.
The camcorder designs we've gotten schematics for all use LCDs with a 505 x 230 pixel arrangement, and thus keep the H1, H2, V1, and V2 control lines 'low' (at ground). You can see, from the horizontal and vertical timing tables in the MN83803A(K) manual (available in our 'Data Sheets' section on this site), how this works.
Also, all set PMODE to 'high'-- something that affects the on-chip VCO frequency and HDATA's rise position (relative to the NTSC sync's onset). We presume, since one of these camcorders uses the MCL0712A01 LCD (and ours is the -A03), that our LCD driver should also use this setting.
All the important control lines are switchable in Julie S. Porter and I's most recent p.c. board design.
For more discussion of the control signals, you may refer to the inferences I've drawn from analysis of the schematic diagrams on our Schematics page.
(2.3) MN83803Ax Series Power Supply Requirements
It is important to note that while we have data sheets for the MN83803AK, some uncertainty arises about the power supply requirements for this and other chips in the MN83803Ax series. The camcorder repair manuals we have obtained show some latitude that we would not expect from the data sheets. We treat this as a system integration issue, in section 6.2, below.
(2.3.1) VDD1 should be 5 volts.
(2.3.2) AVDD should be 5 volts.
(2.3.3) VDD2 should be 18 volts.
(2.4) Passive components used with the MN83803Ax chips
External networks of passive components are best displayed graphically, in schematic diagrams. These can be found on our schematics page. However, where there is some question of how to set up this circuitry, or differences between the schematics, it is useful to summarize these differences. The two networks that in question for the MN83803Ax chips are those that feed pin 5 ('VIN'), and those associated with the RVCO input, pin 11.
(2.4.1) VIN Network
An RC network that appears to implement a bandpass filter functionality feeds into pin 5 on the MN83803Ax, marked 'VIN' on the MN83803AK data sheet and in the camcorder schematics.
Design Scenario #1: The network feeding VIN takes its input from an external sync separator.
In one camcorder series (the PV-D705) this network feeding pin 5 takes its input from the line that feeds pin 6 (SYNC/VIDEO); this is the sync input from the video signal processor (AN2523FAP/K). In this camcorder design, this signal appears to contain negative-going sync pulses, with pin 6 correspondingly set to 'VIDEO' mode. See Julie Porter's tests of the SYNC/VIDEO system's functionality, and Section (2.1), above,for more discussion of the SYNC/VIDEO system. (Click here for a link to the schematic diagram.) In this diagram, the network in question comprises R942, R943, C945, C946, and C947.
Design Scenario #2: The network feeding VIN takes its input from the MN83803Ax's built-in sync separator.
In the PV-950-B and PV-950B-K camcorders this network takes its input from pin 7 of the MN83803AL (SYNC OUT), so in this design the sync separator feeding into the network is the one built into the MN83803Ax, rather than an external sync separator. (Click here for a link to the schematic diagram.) In this diagram, the network in question comprises R941, R942, C952, C953, and C954. This network is identical to the one described above, in the PV-D705 camcorder schematic, except in the value of its input resistor. Its form and the rest of its corresponding component values are the same.
The processing stream inside the MN83803AL that produces the 'SYNCOUT' signal at pin 7, and thus feeds the network leading to pin 5 (VIN), begins at the chip's pin 6 (SYNC/VIDEO), which in this case is set to 'SYNC' functionality, taking a positive-going sync input from the IR3y05y's sync separator. Why this design? Why two sync separators (one on the IR3y05y and one on the MN83803AL)? We just don't have the answer right now.
Discussions of this and other issues are found in the inferences section of our schematics page.
(2.4.2) VCO Network
The MN83803AK/K's VCO network in the PV-D750 uses a different fixed resistor than the PV-950-B and PV-950B-K: Its R937 (4700 ohms) corresponds to the latter two camcorders' R958 (10K ohms)-- and their associated variable resistors are identical (each 10K ohms). The variable resistor in this network controls the centering of the image on the LCD screen.
(2.5) Interchangeability among MN83803A, MN83803AK, MN83803AK/K, and MN83803AL
Serious questions arise from the camcorder schematics about the MN83803 chips' operational power supply voltage range. We consider this as a system integration issue in section (6.1), below.
(3) IR3y05 Series Video Processing Chip
(3.1) NTSC video input
Unmodulated NTSC input (the kind we usually see travelling through RCA connectors on our VCRs, e.g.) drives the IR3y05y; it is 1 volt peak-to-peak, with 0.714 volts above system ground for the video signal, and -0.286 volts below ground for the sync pulse.
(3.2) Video center voltage (DC offset) control
Here's how the DC offset control is set up in the IR3Y05:
The error amplifier's output trims an offset
control network in the R, G, and B output amplifiers (5), keeping the
average voltage of all three channels at the desired voltage offset.
Full-system testing of the IR3y05y has revealed some difficulty in
controlling the RGB output center voltage (DC offset) reliably by
applying external voltage to pin 28. The RGB DC offset sags to between
5.0 and 6.0 volts, typically, no matter what setting is applied to the
control network feeding pin 28.
Since these circuits depart from the ideal in their performance, you might want to feed the 7812 with a trimmer potentiometer's center tap (wiper), fed by an appropriate voltage divider. BUT
BEWARE: a potentiometer's power rating does not apply to its wiper—it
applies to the power flowing through its end terminals—so trimming like
this might end up making the trimmer, and you, more than a little sad,
when the voltage regulator dumps a dose of current out its ground pin,
and fries the wiper. As always, V=IR, and P=I2R!
As you can see in section (3.4),
immediately below, the maximum voltage for the IR3y05's VCC2 is 13.75.
Running a chip at its maximum rated operational voltage is not
conventionally regarded as a good idea-- but we don't know if the
IR3y05y used in camcorders is designed for a different working voltage,
and it appears that the IR3y05 used in the Cybermaxx is fed a voltage
right around its uppermost rated VCC2.
(Technical note: Maximum operational voltages differ from absolute maximum
voltages; we don't expect damage to the chip at maximum operational
voltage like we do at the absolute maximums--but we may reasonably
expect performance to deviate from the ratings given in the data
The default RGB output center voltage is VCC2/2; if this is sufficient,
pin 28 can be left open. If a different center voltage than VCC2/2 is
needed, then an external reference voltage can be applied to pin 28.
See also section 1.1.2 and section 6.2.1, and the IR3y05 manual. Keep the reference voltage within the bounds of VCC2 and ground, to avoid damage to the chip.
You can see here how the DC offset control is supposed
to work. The resistor network (1) feeding pin 28 (2) gives a reference
voltage (the capacitor just shunts off RF interference to ground). An
error amplifier (3) compares this voltage to a sample of the RGB output
(4) (taken from one channel, green), which is time-averaged, with the
help of an external capacitor at pin 35. (You can think of this either
as integration or as low-pass filtering).
DC Offset Problem
Here you can see the equivalent circuit diagram given for pin 28. Can
you find a problem with it? The input protection diodes (at left) look
appropriate, protecting against inputs that go beyond the supply rails
(VCC2 and ground) by more than a diode drop.
Two solutions suggest themselves:
Run a jumper (short) across the positive-most resistor in the network
feeding the control voltage to pin 28. This raises the voltage the
network puts out to pin 28. A preliminary test of this setup suggests
that it doesn't work.;
Run the IR3y05y on a VCC2 voltage of 13.75, as in the camcorders. Doing
so should yield an RGB center voltage just shy of 7 volts, within the
MCL0712A03's specified range. The Cybermaxx runs the IR3y05y's VCC2 at
around 14 volts, like this. The Cybermaxx feeds this voltage to the
chip via a 7812 three-terminal voltage regulator that has its ground
pin held above the system ground by a voltage divider. Ideally, you'd
hold the 7812's ground two volts above system ground to do this. We
suspect that the engineers who built the Cybermaxx used such an odd
power supply configuration for the same reason as we are presently
Here you can see the simple ground shim used to bump up the output voltage from a common 3-terminal 12-volt regulator to 13.75 volts. Just put a voltage divider between a regulated voltage input and ground, calculated so the voltage regulator's ground pin sees 1.75 volts.
I mentioned only two solutions suggesting themselves. Julie Porter,
thinking outside the box (and beyond the data sheets), finds a third:
lower the COM voltage feeding the LCD, so it is 1.7 volts lower than
the unexpectedly low RGB values. Whatever the long-term effect of this,
it puts video on the LCD screen (we are reminded of the American
Pragmatist criterion of truth, which is, simply put, whether something
works or not). See our news on Julie's successful test and a detailed report on Julie's findings for more information.
Here's how the DC offset control is set up in the IR3Y05:
The error amplifier's output trims an offset control network in the R, G, and B output amplifiers (5), keeping the average voltage of all three channels at the desired voltage offset.
Full-system testing of the IR3y05y has revealed some difficulty in controlling the RGB output center voltage (DC offset) reliably by applying external voltage to pin 28. The RGB DC offset sags to between 5.0 and 6.0 volts, typically, no matter what setting is applied to the control network feeding pin 28.
Since these circuits depart from the ideal in their performance, you might want to feed the 7812 with a trimmer potentiometer's center tap (wiper), fed by an appropriate voltage divider. BUT BEWARE: a potentiometer's power rating does not apply to its wiper—it applies to the power flowing through its end terminals—so trimming like this might end up making the trimmer, and you, more than a little sad, when the voltage regulator dumps a dose of current out its ground pin, and fries the wiper. As always, V=IR, and P=I2R!
As you can see in section (3.4), immediately below, the maximum voltage for the IR3y05's VCC2 is 13.75. Running a chip at its maximum rated operational voltage is not conventionally regarded as a good idea-- but we don't know if the IR3y05y used in camcorders is designed for a different working voltage, and it appears that the IR3y05 used in the Cybermaxx is fed a voltage right around its uppermost rated VCC2.
(Technical note: Maximum operational voltages differ from absolute maximum voltages; we don't expect damage to the chip at maximum operational voltage like we do at the absolute maximums--but we may reasonably expect performance to deviate from the ratings given in the data sheets.)
(3.3) Signal inversion control ('frame rate pulse')
This is a TTL-compatible, 5-volt peak-to-peak, positive-going signal. In our circuit it is provided by the MN83803A series chip that drives the LCD.
(3.4) Power supply
The logic sections use a +5 volt supply (VCC1) and the video output sections use a supply from 11.25 to 13.75 volts (VCC2), 12 volts being typical.
(3.5)Passive components used with the IR3y05 and IR3y05y
See the PV-950-B / PV-950-B/K camcorder schematic, and the IR3y05 data sheets, for details on these. A full parts list and parts placement specification are provided for our design, in our PC board section.
(3.6) Interchangeability between IR3y05 series chips
While we have found no deviations from the IR3y05 specifications in the camcorder schematics, we do not know why there is a 'y' appended to the device Panasonic uses in its camcorders. Interchangeability issues are treated in section (6.1) below.
The LCDs need backlighting, shining through them, so the images on them can be seen. This is provided by a special cold-cathode fluorescent tube, sold with the LCDs. You may refer to our backlight page for more information about how to hook up and feed signals to this device.
(4.1) The backlight works on +5 volts and takes a +5 volt, 8-microsecond sync signal during the horizontal blanking interval, roughly corresponding to the 'cathode blanking interval' for a CRT system. (See the data sheets for the Harris CD22402 video sync generator)
(4.2) The MN83803Ax series chips provide this sync for the backlight in camcorder electronic viewfinder video systems.
(5) Physical design
(5.1) Surface Mount Circuit Fabrication .
These devices are extremely small and the traces on the circuit boards that one makes for them will be easy to break. Therefore there is a learning curve involved in working with them. Julie found her -AL chip to be nicely resilient-- but good handling and electrostatic discharge (ESD) precautions are always a good idea-- even if you find yourself working with bipolar devices . CMOS (pervasive now) is not unique in its susceptibility to ESD damage; it just happens to have high input impedance and a breakdown voltage that exacerbates the problem.
A good soldering technique, recommended to me by Nate Caine and documented elsewhere on the web, is to get the chip situated where it belongs by soldering one lead on each corner of it, and then soldering the whole line of pins, not worrying at first about shorts; then sucking up the excess solder with a solder wick. This will get rid of the shorts resulting from solder bridges. You can get solder wick material from Jameco , and other places.
Another method that works well is to pre-tin the printed circuit board traces the I.C. is to mount to with a little extra solder, taking care not to make solder bridges; then to position the chip by soldering first one corner, and then the opposite corner, to their respective printed circuit board traces; and lastly, to solder the remaining leads using a fair amount of pressure to hold the I.C. lead down on the trace it is being soldered to, then relaxing the pressure and drawing the soldering iron away sideways (parallel to the circuit board) so as not to pull the lead up off the trace before the solder cools. Several pins at a time can be soldered in this way, if a 'screwdriver'-type flat tip is used on the iron.
Using this method, I was able to get all but one pin of the MN83803AL soldered to our industrially manufactured P.C. board on the first try, with no solder bridges. The flat-tipped iron worked perfectly. Since the bad connection that remained after the first round was at a corner, I did not have to switch to a narrower tip to repair it, and it was easy to apply solder for the repair. A sharp 1/64" tip is good for repairs at this 0.5mm scale, should they be necessary, though caution is essential so that leads and circuit board traces are not damaged.
The IR3y05y was not so cooperative, but at its 0.75mm pitch, it was easy to make repairs by applying extra solder. The wick technique described above was only necessary once, to remove a solder bridge.
It stands to reason that if the flat-tipped soldering iron can solder several pins at once, a tip designed to heat the whole row of pins could do a whole row at once. I haven't tried this, but such tips are apparently available in desoldering kits, in various configurations matching the dimensions of I.C. packages, such as the Quad Flat Packs (QFPs) the MN83803 and IR3y05 are distributed in.
Remember to follow the data sheets' specifications for temperature and duration of soldering. A variable-temperature solder station with good protection against electrostatic discharge (ESD) is desirable. Generally use 260 degrees centigrade or less, for less than ten seconds at a time. Allow the chip to cool between applications of the iron to the leads. Do not exceed the temperature and soldering-time specification.
(5.1.2) Other SMT resources
We bring together some resources for SMT fabrication at our surface-mount tech page, http://www.gwi.net/~pstewart/smt.html.
The 0.5mm-pitch leads of the MCL0712A03 LCD emerge from a flexible printed circuit (FPC) connector, and mate to connectors that are available from Halted Specialties, Inc -- though Halted (aka HSC) has misplaced the connectors for an extended period in the past, and probably cannot be relied on to find them at any particular time. Molex makes suitable replacements. See our parts sources page for more details.
(6) System Integration
(6.1) I.C. interchangeability questions
We are aware of at least three variants in the MN83803Ax series (MN83803A, MN83803AK, possibly an MN83803AK/K, and MN83803AL), and two in the IR3y05 series of integrated circuits (IR3y05 and IR3y05y).
This would not be a problem if we had complete documentation on all these chips, and if all the chips were available readily and inexpensively from proven-reliable sources. Unfortunately none of these is yet the case. Whether any particular combination of LCD, IR3y05, and MN83803Ax will work together remains to be established, and the documentation we have found so far gives only partial answers to questions of interchangeability and system integration that follow from this uncertainty.
Naming differences within a family of integrated circuits usually reflect differences in functionality, in electrical, thermal or mechanical characteristics, or in packaging. Since the MN83803 series chips are packaged identically, or nearly so, our only concerns need be those of functionality and electrical characteristics (thermal and mechanical characteristics not being likely issues of concern here). The same holds for our IR3y05 series chips.
In the case of the MN83803Ax series chips, there are both functionality and electrical characteristics questions. Because an MN83803AL device shows up in the parts list for one version of the Cybermaxx 180k virtual reality headset, driving an entirely different LCD than ours, we naturally wonder about the LCD configurations it can drive, and the logic control settings appropriate for these. It is listed now as a direct replacement for the MN83803AK, so presumably it uses the same logic settings to drive LCDs of the same configuration as the MN83803AK, but previously it was not listed as a replacement-- or so the Panasonic Services Corporation telephone person I spoke with asserted, quite clearly.
FUNCTIONALITY QUESTIONS IN THE TIMING CAPABILITIES OF THE MN83803A vs. MN83803AK:
Panasonic sells the MN83803A in a box labelled MN83803AK, but the data sheets suggest they are not interchangeable.
Nonetheless, they are sold as if they are interchangeable. We can imagine a variety of scenarios where this might make sense (e.g. a hypothetical case where "AK" chips were mislabeled "A"), but we are left in doubt , which can only be relieved by empirical, in-circuit testing of the chips.
To look at the same issue in a little more detail: As described in section (2.2), above, each chip in the MN83803 series is hard-wired with the ability to drive any of several LCDs, with different pixel configurations and timing requirements. Several logic control inputs to the chip tell it which configuration it is to work for. Our LCD's pixel configuration (505Hx230V) requires all four logic control lines (H1, H2, V1, V2) to be held 'low', at ground. In the column where this is indicated, we find our 505x230 configuration has been pencilled in. It is new in the MN83803AK specification, apparently. This means that some predecessor chip in the MN83803 series, most likely the "A" chip itself, was designed for a different LCD than ours. Presumably that predecessor chip would not work with our LCD. Given the same control logic inputs as the -AK, it would put out different, and inappropriate, timings.
ELECTRICAL CHARACTERISTICS (ESPECIALLY VOLTAGE) QUESTIONS:
There are, similarly, power supply voltage questions. VDD2 sets the drive voltage for the LCD control outputs, and thus is 18 volts for our chip; but in the camcorder manuals we find the -AL running at 14.1 volts for VDD2; an -AK/K running at 18 volts for VDD2; and an -AK running again at 14.1 volts. See table below for more details. With a voltage range of 18 +/- 2 volts given in our MN83803AK manual for VDD2, we again suspect we have incomplete or inaccurate information.
In the case of the IR3y05 series, we have the data sheet for the IR3y05 but none for an IR3y05y; none of the schematic diagrams showing the IR3y05y in situ contradict these data sheets; but questions linger as to why the 'y' variant is so named-- whether indeed it is a variant, and if it is, how.
These considerations taken together introduce considerable uncertainty into the project of designing a driver board using these devices. In time we will resolve these questions empirically-- but for now we are in the dark, and must hope that copying the camcorder designs with fidelity and attention to our LCD's data sheet will produce a working design without expensive trial and error.
(6.2) Power supply questions.
The power supplies of the various devices are a crucial system integration issue. The MN83803Ax's VDD2 must suit the LCD's VDD; the LCD's COM (LCCOM, Vcom) voltage is a function of the RGB video average voltage (center voltage), as well-- which is itself a function of the IR3y05's VCC2 unless a reference voltage is applied at that chip's pin 28. While these amount to almost trivially simple system integration issues when one has complete device data sheets in hand, they add unwanted degrees of freedom to the design-- and all the risk of chip destruction and system malfunction that goes with uncertainty over power supply specifications-- when complete and accurate data sheets are unavailable.
For this reason we provide all the information we find relevant to this issue, to support educated guesses that will, hopefully, reduce the uncertainty in our system design.
The MCL0712A03 LCD is rated for a VDD of 18 volts, as is the output section of the MN83803AK chip that drives it (the supply for which is labeled VDD2 in the chip's data sheets). However, some of the camcorder repair manuals raise questions about VDD2.
The MN83803A(K) data sheets we have (elsewhere on this site) show these chips' operation range goes from 16V (min) to 20V (max), with 18V being typical.
The camcorder schematics and their associated test point voltage markings are a mess, though, making it hard to tell just what is going on with some of the chips. Some of the test point voltages on the +18volt supply line to the LCDs read '0V' and things like this, for example.
You'll see in the table below that both the MN83803AK and MN83803AL chips are shown running on +14 volts VDD2-- contrary to what the former's data sheet says! As a result, we don't know (1) if the MN83803AK is really limited to an 18 +/- 2 volt supply; or (2) whether the MN83803AL will run on 18 volts, or will require 14 volts instead.
For more analysis of this issue, you may refer to the 'inferences' section of our schematics page on this site.
See section (6.2.5), below, for a discussion of the IR3y05 and IR3y05y VCC2 power supplies.
Camcorder Model Designation Video Processor Chip VCC2
Timing Chip VDD2 LCD Type PV-950-B
IR3y05y 13.7V MN83803AL 18.1 V (voltage chart) or
14 V (schematic)
AN2523FAP/K 14.0V MN83803AK 14.1V
IR3y05y 13.7V MN83803AK/K 18.1 V
(6.2.1) IR3y05 supply voltage and the RGB video center voltage
Our LCD's use of a different video center voltage than the one used by the camcorder EVF we are copying necessitates some design changes when applying the IR3y05y video processing chip, since its default RGB center output voltage is 1/2 of VDD2, i.e. 12 volts divided by 2 = 6 volts. This is inappropriate to an LCD requiring pretty close to 7 volts for its RGB center voltage, such as ours does. We hook up an external voltage reference to the IR3y05y chip to control the RGB center voltage it provides, for this reason-- something thoughtfully provided for in the IR3y05y's design (at Pin 28).
See the italicized update appended to Section (3.2), above, for details on our testing of this setup, which suggests setting the output RGB DC offset to VCC2/2 and using a different VCC2 (13.75 volts e.g.).
(6.2.2) IR3y05 VDD1
The logic sections of these chips run on +5 volts.
(6.2.3) Power-on sequence
It is crucial to observe the instructions on the power-on sequence for the separate subsystems on these chips. E.g., powering up the +5volt section of the MN83803Ax series chips first, and then bringing up the +18volt supply second, will destroy them. You need to either bring up the +18volt supply first, or bring it up simultaneously with the +5volt supply. (See MN83803AK data sheets, on the 'Absolute Maximum Ratings' page for details).
(6.2.4) Noise is always an issue. Stick an inductor in series with the line, or a ferrite core around it,, to cut RFI. Capacitively decouple near the chips whenever possible to cut down on power supply noise at the chip.
(6.2.5) IR3y05 & IR3y05y power supplies:
It's unclear if these work on exactly the same voltage range (or how else they may differ), since I haven't been able to locate data sheets for the latter chip. Trying to get the design working for the first time, it's desirable to follow the camcorder schematics-- and use the 'Y' variant of the chip-- but it's easier to find and set up a three-terminal 12-volt regulator for the supply than to make a custom 14-volt (or as in the camcorder schematics, 13.7 volt?)supply-- opening the question of whether the 'Y' variant will really run on 12.0 volts, as the IR3y05 itself will. (Its VCC2 operating range is 11.25 to 13.75 volts).
If the IR3y05y will run safely on 14.0 volts, then it won't need an external reference voltage for its RGB offset adjustment (it would simply set this to VCC2/2 = 7.0 volts). 14-volt zener diodes are easy to find, so running the chip from a custom 14.0-volt supply might not be out of the question, if the chip will safely run at this voltage. If its supply requirements match that of the IR3y05, though, this is out of the question (14 volts would be its absolute maximum VCC2 in that case-- not a good voltage to design for).
Click here to return to my home page.
Click here to return to the top of the present page.
Click here to return to the main LCD project page.